GENERAL TECHNICAL REPORT PSW-GTR-229
Impacts of Phytophthora ramorum on Oaks
and Tanoaks in Marin County, California
Forests Since 20001
Brice A. McPherson,2 David L. Wood,2 Maggi Kelly,2 Sylvia R.
Mori,3 Pavel Svihra,4 Andrew J. Storer,5 and Richard B.
Standiford6
Abstract
The forests of Marin County were among the first in coastal California to be affected by the
Phytophthora ramorum epidemic. Although initially observed in 1994 in tanoaks
(Lithocarpus densiflorus) and 1995 in coast live oaks (Quercus agrifolia), it is evident from
studies of disease progression that the pathogen was present at least several years prior to the
first recorded tree mortality (Rizzo and Garbelotto 2003; McPherson and others 2005). The
causal agent had not been identified in March 2000 when we established 20 plots in two
Marin County sites, China Camp State Park, (CCSP, coast live oaks and California black
oaks, Q. kelloggii) and the Marin Municipal Water District watershed (MMWD, all three
species). Plots were between 320 m2 and 3600 m2, with a mean of 1234 (SE = 199) m2. The
goal of the study was to monitor disease progression (McPherson and others 2000) and, in
particular, to understand the phenomenon of the abundant beetle attacks observed on bleeding
trees. Plots were evaluated four times per year until March 2003, then twice annually
thereafter. Through March 2008, every stem > 5 cm diameter at breast height (DBH) was
evaluated for signs and symptoms of sudden oak death (SOD): bleeding, beetle attacks, and
the presence of the fungus Hypoxylon thouarsianum. In 2001 and 2007, we recorded the basal
area for every woody species with stem DBH > 5 cm found within 0.08-ha subplots placed
within the larger plots.
Mortality of each of the three species has increased steadily since 2000. Calculated infection
and mortality rates were similar for coast live oaks and black oaks, but considerably greater
for tanoaks (table 1). Mortality not attributable to P. ramorum serves as an estimate of
background mortality. The proportion of the total living trees that were symptomatic,
recorded at 1-year increments, increased gradually for coast live oaks and black oaks, whereas
tanoaks exhibited a rapid increase from 2000 to a peak in 2004, then declined. This decline is
likely a result of decreasing numbers of live stems in these heavily impacted sites.
1
A version of this paper was presented at the Fourth Sudden Oak Death Science Symposium, June 1518, 2009, Santa Cruz, California.
2
Dept. of Environmental Science, Policy, and Management, University of California,
Berkeley, CA 94720.
3
USDA Forest Service, Pacific Southwest Research Station, Albany, CA 94710.
4
University of California Cooperative Extension, Novato, CA 94947.
5
School of Forest Resources and Environmental Science, Michigan Technological
University, Houghton, MI 49931.
6
Division of Agriculture and Natural Sciences, University of California, Oakland, CA 94607.
Corresponding author: bmcpherson@berkeley.edu.
210
Proceedings of the Sudden Oak Death Fourth Science Symposium
Table 1—Infection and mortality rates for coast live oak, black oak, and tanoak
from 2000 to 2008 (percent y-1)
Species
Infection
Mortality
Non-P. ramorum Mortality
Coast live oak
5.0
3.1
0.54
Black oak
4.1
2.4
-Tanoak
10.0
5.4
0.75
We modeled the decreasing numbers of asymptomatic coast live oaks and tanoaks resulting
from the disease, using the over-dispersed Poisson regression. A regression line was fitted
using the total number of asymptomatic trees from each plot with time as the independent
variable, from March, 2000 to March, 2008, with plot nested in site as a random effect to
account for the over-dispersion due to the repeated measurements. Asymptomatic coast live
oaks were projected to decrease 50 percent by 2012 and 90 percent by 2036. For tanoaks, a 50
percent decrease occurred by 2005, with a 90 percent decrease projected by 2015.
Infection rates increased for all three species as a function of increasing stem DBH. Beetles
have been shown to preferentially attack the larger bleeding coast live oaks (McPherson and
others 2005). Numbers of attacks increased with increasing canker size (McPherson and
others 2008). As a consequence, the mean size of surviving trees has decreased since 2000.
Disease stage-specific mortality was modeled using Cox proportional hazards (PH) analysis
and Weibull survival analysis (Lee and Wang 2003). For each species, we modeled the
overall survival of all trees that were asymptomatic in 2000. We then applied the analysis to
initially asymptomatic cohorts of trees that 1) developed bleeding, and 2) developed bleeding
and then were colonized by beetles. Variables included site (CCSP or MMWD), plot, and
DBH. Each tree was assessed twice per year, in March and September, from 2000 to 2008.
Survival functions in both models were comparable, with the Weibull curves all lying within
the 95 percent confidence intervals of the corresponding Cox PH curves. Median Weibull
survival estimates (the point where the probability of surviving longer than a given time is
0.5) for trees that were asymptomatic in 2000 were similar for coast live oaks and black oaks,
but less for tanoaks (table 2). For each species, estimates for trees that developed bleeding
were 25 percent to 50 percent lower than the overall estimates. Beetle attacks further reduced
median survival by 65 percent to 80 percent.
Table 2—Estimated median survival times (years) for initially asymptomatic
trees (SE)
1
Bleeding
+ Beetles
Overall
Asymptomatic
Bleeding
CCSP
15.8 (1.5)
11.7 (2.7)
2.4 (0.6);
3.4 (0.8)
MMWD
11.7 (0.8)
7.5 (1.6)
1.9 (0.5);
2.6 (0.5)
Black oak
13.8 (3.0)
6.2 (1.3)
1.9 (0.9)
Tanoak
8.8 (0.7)
5.9 (0.7)
1.7 (0.4)
Species
Coast
live
oak
1
The two values shown for coast live oaks with bleeding and beetles represent cohorts with mean
DBH = 20 cm (upper entry); mean DBH = 40 cm (lower entry).
The survival analyses reported here update earlier Weibull analyses that differed in two
respects. In the earlier model, each same-symptom disease status cohort was composed of
individuals with the same symptom status in March 2000, and trees were evaluated once per
211
GENERAL TECHNICAL REPORT PSW-GTR-229
year (McPherson and others 2005). By 2008, sufficient numbers of trees had progressed from
asymptomatic to bleeding and ultimately to death that we could use only asymptomatic trees
as the starting cohort. The value of the Weibull modeling approach is evident in the
consistency of the estimates derived for symptomatic trees in both the 3- and 8-year time
frames. Although the overall survival estimates for coast live oaks probably overestimated
median survival times in the earlier model, 29.5 (SE = 8.4) y and 31.8 (9.3) y, survival
estimates for bleeding trees were 7.0 (1.2) y and 7.6 (1.6) y, and for beetle-colonized trees, 2.6
(0.3) y and 3.2 (0.6) y, for CCSP and MMWD, respectively (McPherson and others 2005).
For tanoaks, the earlier estimate for overall median survival was 12.6 (3.8) y, for bleeding
trees, 8.7 (2.3) y, and for beetle-colonized trees, 2.9 (1.0) y. In general, the modeled median
survival estimates for symptomatic coast live oaks and tanoaks were more consistent for the
two models than were the estimates for overall survival. Once beetles attacked trees, both
models showed dramatic, consistent decreases in survival.
Basal area of all woody stems was measured in 2001 and 2007 in 0.08 ha plots sited within
the larger plots. Live California bay laurel (Umbellularia californica) basal area increased by
8 percent during this period. The proportion of coast live oaks in the disease progression plots
that were symptomatic and dead was positively related to bay laurel stand basal area in 2001
(F1, 14 = 13.096, P < 0.0028) and in 2007 (F1, 14 = 4.99, P < 0.042). The diminished strength of
the relationship suggests that the role of bay laurel in P. ramorum transmission to coast live
oaks was more important early in the epidemic and decreased with time. Live Pacific madrone
(Arbutus menziesii) basal area decreased 12 percent during the same period. There was a
negative relationship between madrone basal area and the proportion of coast live oaks that
were symptomatic and dead. The strength of this negative relationship increased from 2001
(F1, 14 = 3.41, P > 0.086) to 2007 (F1, 14 = 5.16, P < 0.0028). We also found an inverse
relationship between basal area of bay laurel and madrone in these plots. We hypothesize that
the contrasting trends of the relationship between basal area of bay laurel and madrone is
correlated with moisture availability or with variables dependent on moisture availability,
such as the infection process. In these forests Pacific madrone tends to be found in drier sites
and bay laurel is more dominant in wetter sites.
This study of the responses of individual host trees supports wider use of survival analysis to
model ecological change driven by introduced pathogens. Because relatively few variables are
required to develop predictions of change at the stand and landscape scales, survival analysis
can serve as a useful tool for land management decisions.
Literature Cited
Rizzo, D.M. and Garbelotto, M. 2003. Sudden oak death: endangering California and
Oregon forest ecosystems. Frontiers in Ecology and the Environment. 1(5): 197–204.
Lee, E.T. and Wang, J.W. 2003. Statistical methods for survival data analysis. WileyInterscience, New York.
McPherson, B.A.; Wood, D.L.; Storer, A.J.; Svihra, P.; Rizzo, D.M.; Kelly, N.M. and
Standiford, R.B. 2000. Oak mortality syndrome: sudden death of oaks and tanoaks. Tree
Notes, Number 26, California Department of Forestry and Fire Protection. 6 p.
McPherson, B.A.; Mori, S.M.; Wood, D.L.; Storer, A.J.; Svihra, P.; Kelly, N.M. and
Standiford, R.B. 2005. Sudden oak death in California: disease progression in oaks and
tanoaks. Forest Ecology and Management. 213: 71–89.
McPherson, B.A.; Erbilgin, N.; Wood, D.L.; Svihra, P.; Storer, A.J. and Standiford,
R.B. 2008. Attraction of ambrosia and bark beetles to coast live oaks infected by
Phytophthora ramorum. Agricultural and Forest Entomology. 10: 315–321.
212